The term AI was first coined in 1955 in a proposal for “a 2-month, 10-man study of artificial intelligence be carried out …at Dartmouth College in Hanover, New Hampshire.” In 1968 the world was introduced to HAL, a sentient computer in the film “2001: A Space Odyssey”. In 2011, Watson, a natural-language question-answering computer, competed on “Jeopardy” and defeated two former champions.
In other words, the concept of artificial intelligence has been around for more than 60 years, and every decade the technology improves.
The question becomes, however, is it good enough to drive cars?
In short, the answer is no, not quite yet.
Sure, there have been huge advances in AI supported by the increase in computing power and the development of sensors, however, there’s still a lot of work to be done behind the scenes. The challenge is dealing with the unpredictability of driving. There are so many scenarios that could happen. Furthermore, the technology self-driving cars use to recognize their surroundings, LIDAR, is a really dumb technology. All it does is try to replicate the human eye, with not much better vision or response times. It constantly fails in harsh weather conditions and can easily be sabotaged with a few strips of duct tape across a stop sign.
Most companies who are trying to make self-driving cars also employ hundreds of thousands of workers whose sole job is to teach the autonomous vehicle how to recognize pedestrians, stop signs, and other objects. All this labeling is done by sitting and watching videos frame-by-frame and identifying objects manually. This can lead to human error.
But what about deep learning? Deep learning aims to replicate the human brain, but it still has its faults. Companies like Facebook and Twitter use deep learning to identify bullying and they still haven’t quite gotten it right — even as they have millions of posts and people from which to train their systems.
Even so, most car manufacturers state they will have fully capable self-driving cars by 2020 thanks to continued technological advancements. Toyota even invested 1 billion in AI technologies and robotics just last year. The question still remains if the proper federal laws and regulations will be in place by then as well.
At JCC Bowers, we’re developing technologies that bring artificial intelligence to the car, driver, and the transportation system more broadly. Sign up for our newsletter to learn more about how we are using artificial intelligence to transform transportation.